It’s mid-October. Greenbox is six months past the customer support wall of fire and has just over six thousand subscribers across Perth and Melbourne. Sam is running a small team now. The support emails are triaged. The inbox is under control. She finally has time to look at the data she’s been collecting, and the thing she notices isn’t in any dashboard.
Sam has been manually tagging every email for three months. Not because anyone asked, but because she has a hunch she can’t quite articulate. The tags are simple, the category of question, whether the subscriber is satisfied with the answer, and whether they’ve churned since. She’s been running a little spreadsheet on the side.
On a Tuesday morning she sits down with the spreadsheet and a cup of tea and runs a pivot table. The result stops her.
Subscribers who have ever referred a friend churn at 2.1% per month. Subscribers who have never referred a friend churn at 9.4% per month. Nearly five times the rate.
She knows referrals correlate with love. What she doesn’t know is which direction the causation runs. Do people refer because they love Greenbox, or do they love Greenbox more because they’ve referred someone? Is the act of recommending a product to a friend a cause of retention, not just a marker of it?
She walks into Maya’s office with the spreadsheet and asks the question.
The conversation in Maya’s office
Maya takes a long look at the numbers. Then she says something Sam wasn’t expecting.
“Do you remember what my mother’s recipe swaps used to look like?”
Sam raises an eyebrow. “Your mother’s what?”
“When I was growing up on the farm, my mother was in this informal network of women who swapped recipes and cooking tips. It wasn’t organised. Nobody ran it. It was just a bunch of people who cared about food and wanted to tell each other when something worked. My mother would call her friend Shirley and say ‘I just did a beetroot thing with orange zest, you have to try it.’ And Shirley would tell her mother-in-law. And Shirley’s mother-in-law would tell her bridge club. And by the end of the month, six people had tried the beetroot-and-orange thing.”
“You’re saying we should run a recipe club.”
“I’m saying something bigger. I’m saying our subscribers are already a community. They just don’t know each other. Every week we send them ingredients. Every week they figure out what to do with those ingredients. Every week, thousands of them are inventing small clever solutions to the same problems, what to do with the unexpected kohlrabi, how to make the Tuesday night meal out of whatever’s in the box, which substitutions actually work. And none of them are talking to each other. We’re sitting on the biggest cooking conversation in Western Australia, and it’s all happening in private.”
Sam sets down her tea.
What a community is (and isn’t)
Lee arrives at the office the next day for his usual fortnightly coaching session. Maya tells him about the conversation.
Lee has seen the community-building mistake in a dozen companies. The most common failure mode is treating community like a marketing channel. Build a Slack. Post a few announcements. Offer discount codes. Call it a community. Wonder why nobody engages.
“Community isn’t a place,” Lee says. “It’s a set of relationships between your subscribers, mediated by something they have in common. The thing they have in common is the box. The box arrives on the same day, full of produce they didn’t choose, and they have to figure out what to cook. That’s the shared problem. That’s the substrate for community. You don’t build community on top of your brand. You build community around the problem your subscribers are actually solving.”
Tom, who has wandered in with a coffee, has a question. “So we build a forum?”
“Maybe. Or maybe not. Forums are one way. There are others. The question is: what would make it easier for subscribers to tell each other what they’re cooking this week? And what would make it feel natural, not promotional?”
Three experiments
They decide to run three small experiments over the next six weeks. None of them require building a platform. All of them are designed to see whether subscribers actually want to talk to each other, or whether the conversation only exists in Maya’s imagination.
Experiment one: the recipe postcard. Every week, alongside the box, Greenbox includes a small printed card. On the card: one recipe using three ingredients from this week’s box, written by a subscriber. The subscriber’s first name and suburb are at the bottom. At the top: “Got a recipe you’d like to share? Tag us @greenboxperth with #greenboxkitchen.”
The first week, the recipe is from a subscriber called Katrina who figured out how to pickle the leftover radishes. Sam knows her from the support emails. She asked Katrina in advance, and Katrina was thrilled. The card goes out in 6,234 boxes.
By Thursday evening, Greenbox’s Instagram tag has 41 posts. By Monday, 117. Most of them are subscribers tagging their own versions of the pickled radishes. A few are questions (“should I use apple cider vinegar or white wine vinegar?”). A handful are new recipes that people are offering for future cards.
Experiment two: the weekly cook-along video. Maya, who has never been on camera, reluctantly films a two-minute video each Thursday evening showing what she’s making with this week’s box. The production quality is terrible. The lighting is bad. The videos are unscripted and occasionally include one of her dogs walking across the kitchen. The team posts them to YouTube and links them in the weekly email.
After four weeks, the videos have a combined 3,100 views. The comments section, to everyone’s surprise, becomes the thing. Subscribers start leaving comments like “I made this last night, added a bit of chilli” and “tried this with the fennel from last week’s box, worked great” and “Maya, why do you keep the garlic in the fridge?” Maya reads them all. She starts replying. The replies get their own replies.
Experiment three: the Saturday morning open kitchen. Once a month, Greenbox opens its small packing warehouse to subscribers for two hours on a Saturday morning. Coffee, pastries, and the chance to see the operation. Maya and Sam are there. So are Tom and Priya, who answer technical questions from curious subscribers who want to know how the matching engine works. The first open kitchen has 34 attendees. The second has 62. The third has 89 and they have to apologise for the pastry shortage.
What they discover
Six weeks in, the team has three bits of evidence and a lot of anecdotes. Sam runs another pivot table.
Subscribers who have engaged with any of the three experiments, posted a recipe, commented on a video, attended an open kitchen, have a churn rate of 0.9%. Subscribers who haven’t engaged remain at 9.4%. The engaged cohort is small, about 340 subscribers, or 5% of the base, but it’s growing by about 20-30 subscribers per week.
The causation question is still open. Maybe people engage because they’re about to stay anyway. But the retention numbers are strong enough that Maya makes a decision.
“We’re not treating community as a marketing experiment anymore. It’s part of the product.”
Tom looks at her. “What does that mean operationally?”
“It means Sam gets a community lead role. Not in addition to customer support, as the new version of customer support. Because the best customer support is subscribers helping each other cook the box. It means the recipe cards are a permanent feature. It means the cook-along videos are part of Maya’s job, not a side experiment. It means we do an open kitchen every month and we budget for it. And it means we look at every feature we ship through the lens of ‘does this make it easier or harder for subscribers to talk to each other?’”
The part Lee adds
Lee has one more thing to say before he leaves on Friday.
“There’s a distinction I want you to hold onto. Communities are emergent. You can create the conditions for them, the shared problem, the shared rhythm, the shared artefacts. You cannot create the community itself. The subscribers do that. Your job is to make it easier, not to control it.”
“What happens if it gets out of hand?”
“Define ‘out of hand.’”
Maya thinks. “What if someone posts something I wouldn’t have posted? What if there’s an argument? What if someone starts a thread I don’t like?”
“Then the community is real. If everything that happens feels like marketing you would have written yourself, it isn’t a community. It’s a broadcast. The sign you’ve actually built something is that you start to lose some control over it. The trade-off is worth it, because the same looseness that makes it uncomfortable is what makes it defensible. Nobody can copy the conversation your subscribers are having with each other. A competitor can copy your box. They can’t copy the fact that two hundred people showed up on a Saturday morning to see your warehouse.”
Maya writes down the last sentence. She’d been looking for a way to describe what she was trying to build, and Lee just handed her the words.
The sticky note that replaces “customers”
The next Monday, Maya goes to the whiteboard in the meeting room, where someone had written the word “customers” in green marker during a strategy session back in August. She hasn’t liked the word for a while, but she hadn’t known what to put in its place.
She erases it and writes “subscribers” in red.
Then, underneath, in smaller letters, she writes: “A subscriber is a person. A person who cooks on Tuesday night and again on Friday. A person who didn’t expect the kohlrabi. A person who has a neighbour who might also like a box. We are not in the business of customers. We are in the business of helping people eat well, together, around the thing we deliver.”
Nobody writes a strategy document about this. It’s just on the wall. But over the following weeks, Sam notices that the team has started saying “subscribers” without thinking about it. The language has shifted, and with it, the frame.
Six thousand people are not an audience. They are the community. Greenbox is the thing they have in common. That’s the whole point.